INTRODUCTION

MHC¹ class I molecules bind peptides that are generated in the cytosol by proteasome-mediated degradation of endogenous and foreign proteins (1-3). Proteasomes are multienzyme complexes consisting of a 20 S catalytic core that associates with regulatory subunits to make a 26 S particle. Membrane-permeable proteasome inhibitors prevent the generation of most antigenic peptides and, because of a shortage of peptide, impair MHC class I egress from the endoplasmic reticulum (4-6). The sites of proteasome cleavage can influence the efficiency of antigen processing into MHC class I-associated peptides (7-9). Two MHC-encoded proteasome subunit proteins, LMP-2 and LMP-7, and the proteasome regulator PA28 modify proteasome specificity and enhance MHC class I antigen processing (10-14). Peptides generated in the cytosol by proteasomes are translocated into the endoplasmic reticulum by the transporter associated with antigen processing (15) and, if they conform to the appropriate motif, are bound by MHC class I molecules (16). The assembled complexes travel to the cell surface, where they are presented to CD8+ cytolytic T lymphocytes (CTL).

Cytosolic protein degradation is generally very specific and regulated. Thus, most endogenous proteins contain or acquire degradation signals prior to proteasome-mediated destruction. One of the best characterized mechanisms for marking proteins for cytosolic degradation involves the ubiquitination of target proteins (17). In this pathway, polyubiquitin chains are conjugated to one or more lysine residues of the target protein. Ubiquitin conjugation requires the action of multiple enzymes and is initiated by the recognition of specific protein sequences (17-19). Several signals that promote ubiquitination have been identified and include the N-terminal amino acid (19), internal sequences called "destruction boxes" (20), or even sequences on proteins associated with the degradation substrate (21). Ubiquitin-independent pathways for targeting intracellular protein degradation, possibly involving PEST sequences, have also been described (22-24).

Several lines of evidence suggest a role for the ubiquitin targeting pathway in MHC class I antigen processing. For example, cells with a defect in the E1 ubiquitin activating enzyme are incapable of processing microinjected chicken ovalbumin (25). Additionally, antigens can be targeted for MHC class I antigen processing by modification of the N-terminal amino acid into a destabilizing residue (26). While a correlation between antigen degradation and epitope generation has been demonstrated in multiple studies (6, 26, 27), there are conflicting reports demonstrating that protein degradation rates do not influence antigenicity (28, 29).

We have used macrophage cell lines infected with the intracellular bacterium Listeria monocytogenes as a model to investigate cytosolic antigen degradation and production of MHC class I-presented CTL epitopes. L. monocytogenes enters the cytosol of phagocytic cells by secreting listeriolysin O (LLO), which destroys the phagolysosomal membrane (30). Bacteria multiply intracellularly and secrete proteins that are processed into peptides that are presented by MHC class I molecules (31). In BALB/c (H2^d) mice, LLO is processed into LLO 91-99 and p60, a murein hydrolase, is processed into p60 217-225 and p60 449-457 (32-34). These three peptides are bound by H2-K^d MHC class I molecules and are recognized by L. monocytogenes-specific CTL. Following secretion into the host cell cytosol, p60 is degraded with a half-life of approximately 90 min (35). Approximately 3% of degraded p60 molecules are processed into p60 217-225 (35), and approximately 30% are processed into p60 449-457 (34). LLO is also rapidly degraded in the host cytosol, and approximately 10-20% of degraded LLO molecules give rise to LLO 91-99 (36). Cytosolic degradation of LLO and p60 is proteasome-mediated (6), and there is a tight correlation between 1) the amount of antigen degraded and the number of CTL epitopes (35) and 2) the rate of antigen degradation and the rate of CTL epitope appearance (6).

It is unclear how p60 is targeted for degradation by host cell proteasomes. To determine if p60 contains a protein degradation signal recognized by the host cell, we made deletion mutants that spanned the majority of the mature protein and expressed them in L. monocytogenes. While most regions within p60 contribute to its stability in the host cell cytosol, the carboxyl-terminal region enhances degradation. Substitution of the N-terminal amino acid of p60 with stabilizing and destabilizing residues demonstrates that p60 is an N-end rule substrate. In agreement with our previous findings that degradation of p60 is tightly linked with epitope generation (6), we find a direct correlation between the rate of intracellular degradation of mutant forms of p60 and epitope generation.

EXPERIMENTAL PROCEDURES

L. monocytogenes strain 43251 was obtained from the American Type Culture Collection (ATCC; Rockville, MD) and grown in brain heart infusion medium. Recombinant bacteria strains were grown in medium with spectinomycin (200 µg/ml). P815 mastocytoma cells (DBA/2, H-2^d) and J774 macrophage-like cells (BALB/c, H-2^d) were obtained from the ATCC and cultured in RPMI 1640 (Life Technologies) with 10% fetal calf serum, 2 mM L-glutamine, 50 µM 2-mercaptoethanol, 20 mM HEPES, penicillin, streptomycin, and gentamicin (RP10). CTL clone L9.6 is specific for p60 217-225 in the context of the H2-K^d molecule and was maintained by weekly restimulation with L. monocytogenes-infected J774 cells as described (33).

Synthetic p60 217-225 (KYGVSVQDI) and p60 217Ser-225 (SYGVSVQDI) were purchased from Research Genetics (Huntsville, AL). Peptides were HPLC-purified, and stock solutions were quantified by amino acid analysis. Ac-LLnL-CHO was purchased from Calbiochem and dissolved in Me₂SO. Cycloheximide and anisomycin were obtained from Sigma.

All p60 constructs were generated from an L. monocytogenes-derived p60 gene with a mutation to encode a serine in position 217 as described (34). Point mutations in the codon encoding amino acid 28 and deletions in the gene region encoding amino acids 45-365 were introduced by oligonucleotide-directed in vitro mutagenesis as instructed (Muta-Gene Phagemid in vitro Mutagenesis Kit, version 2; Bio-Rad). The p60 217Ser gene was cloned into phagemid pTZ-19-u and single-stranded DNA was made. The following synthetic oligonucleotides were used to make point mutations: Ser²⁸  Arg, 5-GACTACTACAGTGCGTGCGCTAGCGATTGTTGG-3; p60 Ser²⁸  Asp, 5-GACTACTACAGTGTCTGCGCTAGCGATTGTTGG-3; p60 Ser²⁸  Met, 5-GACTACTACAGTCATTGCGCTAGCGATTGTTGG-3; p60 Ser²⁸  Val, 5-GACTACTACAGTGACTGCGCTAGCGATTGTTGG-3. The following oligonucleotides with in frame deletions contain a restriction site for NheI, encoded by silent mutations in the codons for amino acid 25 and 26: p60 45-90, 5-GCGCCACTACGGACGTTACTTTGTGCGATACCCC-3; p60 91-141, 5-GGAGTGCTTACTGCTTTGTCTAACCAAGTTGCGCT-3; p60 138-179, 5-GGCGCAGGTGTAGTTGCACCGTTAACGAAACCAG-3; p60 170-216, 5-CTTGAACAGAAACACCGTAGCTTTCTGCAGCAGGTGCAGC-3; p60 226-266, 5-AACTACTGGAGCTGCTTGAATGTCTTGAACAGAAAC-3; p60 265-305, 5-GTAGATGGTGCAGGAGCAGCTGCTGGAGCTTCCGT-3; p60 304-365, 5-AGCACTTGCACTTGAATTTGCAGCTTCTGTTGGTGC-3.

Carboxyl-terminal deletions (358-484) were introduced by the polymerase chain reaction using p60 217Ser and p60 Arg²⁸, p60 Asp²⁸, p60 Met²⁸, and p60 Val²⁸ 217Ser point mutants as templates. The 5 oligonucleotide (5-GGGTCGACTCGATCATCATAATTCTGTCTC-3) encodes a SalI restriction site and includes 442 base pairs upstream of the p60 translation initiation site. The 3 oligonucleotide (5-GGAAGAACCTTAATTAGCATTCGT-3) introduces a stop codon at p60 amino acid 358. Polymerase chain reaction products were inserted into the TA vector (Invitrogen, San Diego, CA). All constructs were verified by dideoxy sequencing. Mutant p60 genes were cloned between the SalI and EcoRI sites of the shuttle vector pAT29, transformed into Escherichia coli HB101 (pRK24) and conjugated into L. monocytogenes 43251 as described (35).

L. monocytogenes were grown overnight to stationary phase. Cultures were centrifuged to remove bacteria and separated by SDS-PAGE (35). Proteins were transferred electrophoretically to nitrocellulose, and membranes were blocked with 5% (w/v) dry milk, 0.1% Tween 20 and probed with anti-p60 rabbit antiserum at a 1:4000 dilution (35). Blots were incubated with horseradish peroxidase-conjugated anti-rabbit immunoglobulin (IgG) in the secondary step and developed by enhanced chemoluminescence (Amersham, Buckinghamshire, United Kingdom).

L. monocytogenes-infected J774 cells were metabolically labeled as described previously (35). In short, 4 × 10⁶ J774 cells were seeded in flasks and incubated overnight in RP10 medium with 100 µg/ml spectinomycin as the only antibiotic (RP-spec). Cells were infected with log phase cultures of recombinant L. monocytogenes (A₆₀₀ = 0.1) for 30 min, washed, and incubated in RP-spec for 3 h. Gentamicin (50 µg/ml) was added after the first 30 min to inhibit extracellular bacterial growth. J774 cells then were washed and placed in methionine-free medium (methionine-free Dulbecco's modified Eagle's medium with 3% dialyzed fetal calf serum, 2 mM L-glutamine, 20 mM HEPES, pH 7.5) with spectinomycin, gentamicin, the eukaryotic protein synthesis inhibitors cycloheximide (50 µg/ml) and anisomycin (30 µg/ml), and 25 µM calpain inhibitor I (LLnL) as indicated in the figure legends. After 30 min, translabel [³⁵S]methionine at a specific activity of 200-500 µCi was added, and the cells were pulsed for periods varying from 20 min to 1 h. Cells were washed and chased for the time intervals indicated in the figure legends in RP10 with 20 µg/ml tetracycline to inhibit further intracellular bacterial growth. Cells were harvested in 1% Triton X-100 lysis buffer with protease inhibitors, and the detergent lysates were cleared by centrifugation (8 min at 14,000 rpm) at 4 °C. p60 was immunoprecipitated with 25 µl of protein A-Sepharose (50% slurry) and 5 µl of anti-p60 antiserum for 1 h at 4 °C. Beads were washed four times and resuspended in sample buffer, and the samples were electrophoresed on SDS-10% polyacrylamide gels under reducing conditions. Gels were enhanced with 0.5 M salicylate, 3% glycerol, dried, and exposed for autoradiography. Radioactive signals were quantified using a Bio-Rad GS-250 Molecular Imager.

Recombinant L. monocytogenes were grown to log phase in brain heart infusion medium with 200 µg/ml spectinomycin. Bacteria from a 1-ml culture were pelleted, washed, and resuspended in methionine-free medium. After 30 min at 37 °C, bacteria were pulse-labeled with 100 µCi of [³⁵S]methionine for 20 min, washed, and chased for 0, 30, and 60 min in RP10 with tetracycline. p60 was immunoprecipitated from the pulse medium and chase medium as described above. Samples were separated by SDS-PAGE, and gels were exposed for autoradiography and PhosphorImager analysis.

CTL epitopes were extracted and HPLC-purified as described previously (33). In short, J774 cells were grown to confluence in 150-cm² plates in RP10 with spectinomycin added the night before infection. Cells were infected with log phase cultures of recombinant L. monocytogenes for 30 min. The medium was then replaced with RP10 containing 5 µg/ml gentamicin, and the cells were cultured for 6 h and then harvested in phosphate-buffered saline and pelleted. Cell pellets were stored at 80 °C. To elute peptides, the cells were resuspended in 10 ml of 0.1% trifluoroacetic acid, Dounce-homogenized, and sonicated. Lysates were ultracentrifuged at 100,000 × g for 35 min, and the resulting supernatants were lyophilized. Protein pellets were resuspended in 0.1% trifluoroacetic acid, passed through Centricon-10 membranes, and separated by reverse phase HPLC. Fractions were collected, lyophilized, resuspended in 200 µl of phosphate-buffered saline, and stored at 20 °C.

CTL assays were performed as described previously (33). Briefly, 10^{4 51}Cr sodium chromate-labeled P815 target cells were incubated with 50 µl of the HPLC-separated peptides and assayed for recognition by p60 217-225-specific CTL clone L9.6 (33). Using our standard protocol, p60 217-225 and p60 217Ser-225 elute in distinct fractions (34). p60 peptide-containing fractions were titrated on P815 target cells and assayed in triplicate along with standard concentrations of synthetic p60 217-225 and p60 217Ser-225. Molar amounts of peptide were determined, and epitope numbers per cell were calculated as described previously (33).

RESULTS

To investigate how L. monocytogenes p60 is targeted for cytosolic degradation and processing into MHC class I-associated peptides, p60 gene deletion mutants were constructed (Fig. 1). Contiguous sections of p60 were deleted by site-directed mutagenesis from a gene encoding p60 217Ser. We used p60 217Ser because p60 217-225 generated from this protein contains serine instead of the wild type lysine in position 217, which alters the HPLC elution time of the epitope without affecting its recognition by the CTL clone L9.6 (34). The p60 signal sequence (amino acids 1-27) was maintained in all p60 constructs to ensure bacterial secretion (Fig. 1B). The deleted and truncated p60 genes were cloned into pAT29, a plasmid that replicates in Gram-negative and positive bacteria, and then introduced into L. monocytogenes. The resulting recombinant strains express both wild type p60, encoded by the chromosomal gene copy, and a plasmid-encoded, mutated form of p60. All recombinant strains multiplied at rates comparable with that of wild type L. monocytogenes and had a normal morphology (not shown), indicating that mutant forms of p60 did not interfere with bacterial growth and septation. To determine the rate of mutant p60 secretion, the supernatants of transformed L. monocytogenes cultures were separated by SDS-PAGE and probed in a Western blot with a polyclonal p60-specific antiserum (Fig. 2). The rate of mutant p60 secretion varied depending on which region of p60 was deleted. The amount of secreted, mutant p60 was proportional to the amount remaining associated the bacterial pellets (results not shown). p60 mutants lacking the N-terminal region (28-45) or the p60 217-225 epitope encoding region were not secreted at detectable levels and therefore were not included in our analyses. Altogether, approximately 94% of the p60 protein was deleted in this panel of p60 mutants.

Wild type p60 is degraded in the cytosol of infected cells with a half-life of approximately 90 min (35). To investigate the intracellular degradation of the deleted and truncated forms of p60, J774 macrophage-like cells were infected with recombinant L. monocytogenes. Infected cells were metabolically labeled and chased for increasing time intervals, and p60 was immunoprecipitated and subjected to SDS-PAGE, autoradiography, and PhosphorImager analysis. Confirming our previous results, we found that wild type p60 secreted by the recombinant L. monocytogenes strains was degraded in the cytosol with a half-life of 60-90 min (Fig. 3). In contrast, deleted p60 138-179, 265-305, and 304-365, was degraded much more rapidly (Fig. 3, A-C). Further experiments established half-lives of 15-20 min for these p60 products (Fig. 3E, Table I). In marked contrast to deletions in the N terminus or midsection of p60, deletion of amino acids 358-484 of p60 (p60 358-484) produced a modest increase in the stability of p60 in the host cell cytosol (Fig. 3D).

Table I. Intracellular half-lives of p60 variants p60 mutants Intracellular half-lifea min Wild type p60 60-90 p60 217Ser   Full-length NDb   45-90c 23   91-141c 18   138-178 22   170-216c 50   226-266c 15   265-305 16   304-365 15   358-484 92-140   Val28 358-484 180   Asp28 358-484c 17   Met28 358-484 84 a Infected J774 cells were subjected to pulse-chase analysis (see legend to Fig. 3). p60 bands were quantified from the polyacrylamide gels using a PhosphorImager, and half-lives were determined by plotting percentages of remaining p60 against the chase time intervals. b Not determinable, because this mutant cannot be distinguished from wild type p60. c Pulse labeling was in the presence and chase in the absence of LLnL to enable half-life determination.

LLnL, a membrane-permeable inhibitor of proteasomes, inhibits the proteolysis of p60 in infected J774 cells (6). We found that degradation of each p60 mutant was abrogated by treatment with LLnL (Fig. 3E and results not shown), suggesting that their degradation is also proteasome-mediated. Since LLnL enhances the metabolic labeling of intracellular L. monocytogenes (36) and since LLnL-induced effects on cellular proteolysis are rapidly reversible (4), we decided to use this inhibitor to facilitate the detection and analysis of poorly secreted p60 mutants. When J774 cells were pulse-labeled in the presence of LLnL and chased in its absence, wild type p60 was degraded normally (results not shown). Similar analysis of J774 cells infected with L. monocytogenes strains p60 45-90, p60 91-141, p60170-216, and p60 226-266 revealed that they were rapidly degraded with half-lives of 15-50 min (Table I). In summary, we find that deletions throughout the N-terminal and middle regions of p60 increase the intracellular degradation rate of this antigen, while truncation of the C terminus is modestly stabilizing.

Although most deletions in p60 affected its intracellular stability, none resulted in a markedly prolonged half-life. Since the N-terminal residue determines the degradation rate of certain proteins (17-19), we decided to mutate the N terminus of mature p60 and to examine the effect on degradation in the host cell cytosol. Like most secreted bacterial proteins, p60 contains a signal sequence that is cleaved upon secretion. The specificity of signal proteases is dictated by amino acids within the signal sequence, particularly amino acids in the 1- and 3-positions. The protein sequence downstream from the signal sequence cleavage site, with the exception of proline in the +1-position, does not appear to significantly influence signal sequence cleavage (38). In the case of p60, the signal sequence is 27 amino acids long, and mature p60 has the amino acid 28 serine at its N terminus (37, 39). We mutated the codon for amino acid 28 from serine to the stabilizing amino acids valine and methionine and to the destabilizing amino acids aspartic acid and arginine (Fig. 1B). These mutations were performed in full-length p60 and also in p60 358-484, which can be discriminated from the wild type form by SDS-PAGE (Fig. 1B). To establish that these mutants were secreted, we pulse-labeled extracellularly grown recombinant bacteria and chased in medium with tetracycline, added to stop bacterial protein synthesis. p60 was immunoprecipitated from the culture supernatant at the time intervals specified in Fig. 4. Wild type p60 was detected in the pulse media but not in chase media, indicating that bacterial p60 synthesis and secretion are essentially concurrent. In contrast, secretion of C-terminal truncated p60s was slightly delayed, with 10% of p60 358-484 and 30% of the p60 Val²⁸ 358-484 and p60 Met²⁸ 358-484 still being secreted during the first 30 min after pulse labeling. In the case of p60 Asp²⁸ 358-484, roughly 50% was secreted during the first 30-min chase period. p60 Arg²⁸ 358-484 (not shown) was secreted at levels too low for meaningful quantitation and could therefore not be used in further experiments.

Secretion of N-modified p60 358-484 p60 by recombinant L. monocytogenes.

We next compared the intracellular stability of the p60 Val²⁸ 358-484, p60 Asp²⁸ 358-484, and p60 Met²⁸ 358-484 mutants with that of p60 358-484 by pulse-chase analyses (Fig. 5). As shown previously in Fig. 3, we again found that the rate of p60 358-484 degradation was slightly diminished in comparison with wild type p60 (Fig. 5A). In contrast, the amount of p60 Val²⁸ 358-484 continued to increase during the first 60 min of chase (Fig. 5B), probably resulting from delayed secretion of this mutant form. During later chase intervals, however, very little degradation was observed, indicating that p60 Val²⁸ 358-484 is remarkably stable in the host cell cytosol. The calculated half-life between 60 and 180 min of chase was 3 h, while, in comparison, full-length wild type p60 was degraded with a half-life of 90 min (Fig. 5B).

The N-terminal residue influences the rate of p60 358-484 degradation.

Like p60 Val²⁸, p60 Met²⁸ 358-484 accumulated intracellularly in the first hour after pulse labeling (Fig. 5D). Degradation of this mutant form during later chase periods was more rapid, with a t1/2 of 84 min between 60 and 180 min of chase. Therefore, placement of a Met at the N terminus does not have the same stabilizing effect as a Val in this position. Although the initial accumulation of p60 Met²⁸ 358-484 during the early chase period may be explained by delayed secretion, an alternative possibility is that p60 Met²⁸ undergoes time-dependent modification prior to degradation. For example, the (³⁵S-labeled) methionine 28 may be removed by cytosolic Met aminopeptidases (40), exposing the position 29 threonine, a destabilizing residue, at the N terminus of the protein.

To facilitate analysis of p60 Asp²⁸ 358-484, infected J774 cells were treated with LLnL during the pulse and initial 60-min chase period, to provide an opportunity for p60 Asp²⁸ 358-484 to be secreted into the host cell cytosol. p60 was then immunoprecipitated immediately and 15 and 45 min after the removal of LLnL. Fig. 5C demonstrates that p60 Asp²⁸ 358-484 was rapidly degraded, with an approximate half-life of 17 min.

Our experiments indicate that the identity of the N-terminal amino acid of mature, C-terminal truncated p60 plays a significant role in its cytosolic stability. Whereas Asp²⁸ is clearly destabilizing, Met²⁸ and the wild type Ser²⁸ confer intermediate stabilities, while Val²⁸ is the most stabilizing residue.

Generation of p60 217-225 requires p60 degradation and is directly related to the intracellular p60 concentration (6, 35). To investigate whether the rate of mutant p60 degradation correlates with the rate of epitope production, J774 cells were infected with recombinant bacteria, and CTL epitopes were extracted, HPLC-purified, and quantified. HPLC fractions were assayed with CTL clone L9.6 for the presence of p60 217-225 and p60 217Ser-225 epitopes, which elute in different HPLC fractions (34). This allowed us to discriminate between epitopes produced from wild type or mutant forms of p60. Epitopes in targeting fractions were quantified by titration of the appropriate HPLC fractions and comparison with a standard curve obtained with precisely quantified synthetic p60 217-225 and p60 217Ser-225, as described previously (33). We found that p60 217-225 numbers were comparable for cells infected with the different recombinant strains (Table II), indicating similar degrees of infection and intracellular bacterial growth. To facilitate the comparison of epitope production from the different deleted forms of p60, the numbers of extracted p60 217Ser-225 epitopes were normalized to compensate for variations in p60 secretion (Table II). The largest numbers of p60 217Ser-225 epitopes were generated from p60 mutants with internal deletions, which correlates with their high rate of intracellular degradation. However, the large number of p60 217Ser-225 epitopes obtained from several of the constructs (226-266, 265-305, 304-365, and especially 45-90) is too great to be accounted for by enhanced degradation alone. In these cases it is likely that the deletions increase the efficiency of epitope generation.

Table II. CTL epitope production in J774 cells infected with recombinant L. monocytogenes p60 mutants Secretiona Epitopes/infected cellb Normalized numbersc p60 217-225 p60 217Ser-225 % Wild type p60 2716 (501) p60 217Ser   Full-length NDd 1236 (55) 9391 (1049)   Val28 ND d 1284 (338) 1798 (112)   Asp28 ND d 1573 (168) 167 (49)   Met28 ND d 1316 (200) 4441 (403)    45-90 4 1758 (241) 674 (21) 16,850   91-141 5 1180 (146) 241 (104) 4820   138-178 6 1975 (221) 217 (63) 3617   170-216 3 843 (42) 96 (42) 3200   226-266 4 1336 (272) 385 (21) 9625   265-305 13 1541 (481) 1038 (78) 7985   304-365 80 1284 (723) 6421 (1003) 8026   358-484 300e 1300 (97) 6117 (250) 2039   Val28 358-484 300c 1753 (406) 395 (27) 132   Asp28 358-484 30 2032 (109) 809 (87) 2697   Met28 358-484 300c 1676 (58) 1079 (33) 360 a The relative amount of mutant p60 is expressed as the percentage of wild type p60 secreted by recombinant L. monocytogenes in infected cells. These values were determined by metabolic labeling of infected cells in the presence of LLnL and separation of immunoprecipitated p60 by SDS-PAGE followed by quantitation by phosphor image analysis (see legend to Fig. 5). b J774 cells were infected with recombinant bacteria for 6 h and then acid-extracted. Eluted peptides were HPLC-separated, and fractions were tested for recognition of p60 217-225 (HPLC fractions 41 and 42) and p60 217Ser-225 (HPLC fractions 43 and 44) by specific CTL clone L9.6. Peptides in targeting fractions were quantified, and the mean epitope numbers per infected cell were calculated from triplicate experiments as described (33). S.D. values are shown in parentheses. c In order to facilitate the comparison of the different p60 mutants, we corrected for their different rates of secretion by normalizing the p60 217Ser-225 epitope numbers. The numbers listed in the last column of this table are the calculated p60 217Ser-225 epitope numbers if mutant p60 secretion occurred at 100% the rate of wild type p60 secretion. d ND, not determined (wild type p60 and full-length mutant forms cannot be separated by gel analysis). e p60 secretion was quantified from the culture media of metabolically labeled extracellular bacteria (see Fig. 4).

Changing the N-terminal amino acid of mature p60 dramatically affected the generation of p60 217Ser-225. Because of unknown secretion levels, it was not possible to compare the epitope yields from full-length p60 mutants in infected cells (Table II). However, nearly 1800 and 4400 p60 217Ser-225 epitopes were generated from p60 Val²⁸ and p60 Met²⁸, respectively, reflecting the relative rates of intracellular secretion and degradation of these mutants. In the case of the N-modified, 358-484 truncated forms of p60, it was possible to correlate epitope generation with antigen secretion and degradation. After correcting for quantitative differences in antigen secretion, we found that roughly 20-fold fewer epitopes were produced from slowly degraded p60 Val²⁸ 358-484 and 7-fold fewer epitopes from the intermediately degraded p60 Met²⁸ 358-484 than from p60 Asp²⁸ 358-484. Epitope production from p60 358-484 was relatively high in comparison with p60 Met²⁸ 358-484, perhaps reflecting more rapid secretion of the former antigen. Alternatively, the relatively rapid degradation of p60 Met²⁸ 358-484, as detected in pulse-chase analyses (Fig. 5D), may reflect principally removal of the N-terminal methionine (which will reduce the ³⁵S signal by nearly 50% since there is only one other methionine and one cysteine in p60 358-484) rather than complete degradation of the truncated p60. Taking these factors into account, our findings demonstrate the direct linkage between antigen degradation and CTL epitope generation.

DISCUSSION

Cytosolic antigen degradation is fundamental to the generation of most MHC class I-presented peptides. In this report we have used an intracellular bacterial model to investigate cytosolic targeting of antigens for degradation and the relationship between degradation and epitope generation. In our analysis, we deleted sequential regions of the secreted L. monocytogenes p60 protein and determined the impact of these mutations on intracellular stability. Deletion of 40-60-amino acid-long stretches from the N-terminal and middle region markedly enhanced p60 degradation. One possible explanation for this finding is that these deletions impair the folding of p60 following bacterial secretion, in which case mutant p60 may have the appearance of a denatured protein. Denatured proteins have been shown in reticulocyte lysates to be good substrates for ubiquitination and subsequent degradation (41). It is noteworthy that most of these short lived p60 mutants have only low expression levels in L. monocytogenes, perhaps resulting from rapid intrabacterial degradation. Thus, the proteolytic pathways that degrade aberrant proteins in bacteria and eukaryotes may respond to similar signals.

Protein degradation in the cytosol of eukaryotic cells is very specific. One mechanism that determines the stability of cytosolic proteins involves the identity of the protein's N-terminal amino acid and is referred to as the N-end rule. Destabilizing and stabilizing amino acids have been determined in yeast, bacteria, and mammalian cells (19). In mammalian reticulocyte extracts, aspartic acid, arginine, and serine are destabilizing residues, while valine and methionine are stabilizing (42). The same hierarchy exists in mouse L cells, with the exception that N-terminal Ser is a stabilizing rather than a destabilizing residue (43). While the N-end rule has been defined using model proteins as degradation substrates, the number of known physiological substrates remains very small (17, 19). Furthermore, the role of the N-end rule in antigen processing is controversial. While one report has shown that placement of a destabilizing residue at the N terminus of an antigen enhances the presentation of CTL epitopes (26), another report has demonstrated that epitopes are presented equivalently from rapidly degraded or stable antigens (29).

We took advantage of the fact that p60 secretion into the host cell cytosol is accompanied by cleavage of a signal peptide, thus exposing a new N terminus (37, 39). This allowed us to investigate the role to the N terminus of mature p60 on its degradation in the host cell cytosol. We mutagenized the wild type serine N terminus of mature p60 into aspartic acid, valine, and methionine. To fully demonstrate the impact of the N-terminal amino acid on antigen stabilization, mutation of the N terminus was combined with deletion of the C terminus of p60 358-484, which by itself has only a modest effect on the intracellular stability of p60. We found that aspartic acid at the N terminus enhanced p60 degradation. A valine at the N terminus stabilized p60. These findings suggest that p60 is an N-end rule substrate. While methionine at the N terminus did not substantially alter the intracellular half-life of p60 in pulse-chase analyses (Fig. 4D) when compared with p60 with wild type serine in this position, the number of epitopes produced from either full-length or C-terminally truncated p60 Met²⁸ was significantly diminished (Table II). This suggests that p60 Met²⁸ is stabilized relative to p60 Ser²⁸ (wild type) and that the rapid degradation we determine may reflect removal of methionine from the N terminus rather than complete, rapid degradation of the whole antigen. Remarkably, while aspartic acid at the N terminus led to very rapid degradation, with a half-life in the cytosol of 17 min, stabilization by N-terminal valine was only partial, with a prolongation of the half-life to 180 min. Thus, it is clear that alternative degradation signals exist for the recognition of p60.

Given the discrepancy in stabilizing effect of N-terminal Ser in mammalian reticulocyte lysates (42) versus in mouse L cells (43), it is interesting that in our system the stability of p60 358-484 with Ser at the N terminus falls between mutants with N-terminal Val and Asp. As discussed earlier by Varshavsky and co-workers (43), the observed differences may reflect a difference in the acetylation of N-terminal Ser under the varying conditions. Acetyl conjugation (which blocks recognition by the N-end rule pathway) may occur more readily onto the the N-terminal Ser of an endogenously synthesized substrate in mouse L cells than in reticulocyte lysates. Although postsecretion modifications of the N-terminal amino acid of mutant forms of p60 may influence their subsequent degradation rates, our findings indicate that differences in the identity of the N-terminal amino acid influence p60 stability in a fashion that, overall, is consistent with the N-end rule, as defined in reticulocyte lysates and in L cells.

The rate of p60 degradation, as determined by the N-terminal amino acid, strongly influences the rate of CTL epitope generation. We previously described the relationship between intracellular antigen concentration, stability, and the efficiency of CTL epitope production (35). Significant change in the rate of antigen degradation, such as changing the half-life from 180 to 15 min, increases epitope production approximately 330%. We find, however, a nearly 20-fold decrease in p60 217Ser-225 generation from p60 Val²⁸ 358-484 (t1/2 = 3 h) compared with the short lived Asp²⁸ 358-484 p60 mutant. This suggests that the proteolytic pathway that degrades p60 Val²⁸ 358-484 is not only slower but is significantly less efficient at generating the CTL epitope.

Our findings extend our understanding of the role of degradation signals in antigen processing and are in agreement with previous studies on N-end rule substrates (26), which demonstrated linkage between antigen degradation and epitope generation. Superficially, our findings appear to disagree with the work of Shastri's and co-workers (29), who showed that the efficiency of antigen presentation did not correlate with the rate of antigen degradation. However, our systems are rather different in that we are investigating epitope generation from antigen introduced into the cytosol, while they are investigating epitope generation from antigen synthesized in the cytosol. Thus, as recently reviewed by Yewdell et al. (44), the majority of CTL epitopes in the latter case may be derived from defective ribosomal products rather than resulting from the actual degradation of the mature protein substrate. In contrast, our experimental system focuses on epitope generation from intact proteins.

Our study suggests that multiple proteolytic signals can target proteins into the MHC class I antigen processing pathway. In our analysis of the p60 deletions mutants, we found that epitopes were generated from all forms of p60 (Table II), regardless of the rate of degradation. The direct correlation between degradation rates and epitope production indicates that recognition and destruction of cytosolic p60 by the host cell proteolytic machinery is the limiting step in CTL epitope generation. Metabolic stability of pathogen-derived proteins therefore is an important factor influencing antigenicity. Indeed, the rate of bacterial antigen degradation may be a more important determinant of antigenicity than antigen prevalence.